Here I want to apply the projected neighbors graph visualization to the neural crest dataset that is used in the scVelo demo.

Setup and get data from scVelo

Use the reticulate package to use scVelo from within R:

Compute velocities on neuro data using velocyto

Extract spliced and unspliced data

Filter genes

Downsample cells to make things easier

Normalize for dimensional reduction

## Warning in if (!class(counts) %in% c("dgCMatrix", "dgTMatrix")) {: the condition
## has length > 1 and only the first element will be used
## Converting to sparse matrix ...
## Normalizing matrix with 977 cells and 4589 genes

Dimensional reduction

Run velocyto on neuro data

Graph visualization

Scores of observed and projected states in PC space

Graph visualization on subset of cells from PC coordinates

Graph visualization on subset of cells from gene expression
using common.genes (intersect of overdispersed genes, odsGenes, and genes in velocity output (genes with high correlation b/w spliced and unspliced))

Graph parameters

Effects of changing k, distance measure, similarity measure, and similarity threshold:
Using PC generated graph

L1 vs L2 as distance measure:

#using k=10, similarity=cosine, threshold=0.25
k = 10
t = 0
set.seed(1)
graphViz(curr.sub,proj.sub,k,"L1","cosine",t,weighted = TRUE, cell.colors = cell.cols.grph,title = paste("L1 distance,K =",k,"weighted, thresh =",t))

set.seed(1)
graphViz(curr.sub,proj.sub,k,"L2","cosine",t,weighted = TRUE, cell.colors = cell.cols.grph,title = paste("L2 distance,K =",k,"weighted, thresh =",t))

Pearson correlation vs Cosine similarity:

k = 10
t = -1
set.seed(1)
graphViz(curr.sub,proj.sub,k,"L2","cosine",t,weighted = TRUE, cell.colors = cell.cols.grph,title = paste("Cosine similarity,K =",k,"\nweighted, thresh =",t))

set.seed(1)
graphViz(curr.sub,proj.sub,k,"L2","pearson",t,weighted = TRUE, cell.colors = cell.cols.grph,title = paste("Pearson correlation,K =",k,"\nweighted, thresh =",t))

..looks like correlation is more conservative than cosine similarity.

Number of out edges k:

Similarity threshold:

## [1] -1
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.1061219  min.arrow.size= 0.002122438  max.grid.arrow.length= 0.0610458  done
## [1] 0
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.1061219  min.arrow.size= 0.002122438  max.grid.arrow.length= 0.0610458  done
## [1] 0.25
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.1173465  min.arrow.size= 0.002346931  max.grid.arrow.length= 0.0610458  done
## [1] 0.5
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.09793795  min.arrow.size= 0.001958759  max.grid.arrow.length= 0.0610458  done
## [1] 0.7
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.09157006  min.arrow.size= 0.001831401  max.grid.arrow.length= 0.0610458  done
## [1] 0.9
## [1] "Done finding neighbors"
## [1] "Done making graph"

## delta projections ... sqrt knn ... transition probs ... done
## calculating arrows ... done
## grid estimates ... grid.sd= 0.09084025  min.arrow.size= 0.001816805  max.grid.arrow.length= 0.0610458  done

Graph parameters consistency scores

Number of out edges k:
L2 distance, cosine similarity, similarity threshold = 0

Similarity threshold:

Consistency of fdg compared to other embeddings

Consistency score in FDG compared to PCA and UMAP computed on same cell subset

## Warning in vattrs[[name]][index] <- value: number of items to replace is not a
## multiple of replacement length

## [1] "Mean consistency scores for PCA, UMAP, FDG"
## [1] 0.2308163
## [1] 0.2293477
## [1] 0.2097686
## [1] "Median consistency scores for PCA, UMAP, FDG"
## [1] 0.188614
## [1] 0.1907555
## [1] 0.170498